infong
/
php-readability
mirror of https://github.com/j0k3r/php-readability
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Merge pull request #14 from j0k3r/revert-bc

Browse Source 
Revert BC changes
pull/15/head 1.1.1
Jeremy Benoist 10 years ago
parent 7b47e2f1de 00f622e9b7
commit dec4514c00
2 changed files with 59 additions and 32 deletions
Whitespace
Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL
Unified View
Diff Options
Show Stats Download Patch File Download Diff File
  1. 62
      src/Readability.php
  2. 29
      tests/ReadabilityTest.php

62
src/Readability.php
Unescape View File

@ -174,14 +174,15 @@ class Readability implements LoggerAwareInterface
* @param string (optional) Which parser to use for turning raw HTML into a DOMDocument * @param string (optional) Which parser to use for turning raw HTML into a DOMDocument
* @param bool (optional) Use tidy * @param bool (optional) Use tidy
*/ */
public function __construct($html, $url = null, $parser = 'libxml', $useTidy = true) public function __construct($html, $url = null, $parser = 'libxml', $use_tidy = true)
{ {
$this->url = $url; $this->url = $url;
$this->html = $html; $this->html = $html;
$this->parser = $parser; $this->parser = $parser;
$this->useTidy = $useTidy && function_exists('tidy_parse_string'); $this->useTidy = $use_tidy && function_exists('tidy_parse_string');
$this->logger = new NullLogger(); $this->logger = new NullLogger();
$this->loadHtml();
} }
public function setLogger(LoggerInterface $logger) public function setLogger(LoggerInterface $logger)
@ -235,6 +236,8 @@ class Readability implements LoggerAwareInterface
* Load HTML in a DOMDocument. * Load HTML in a DOMDocument.
* Apply Pre filters * Apply Pre filters
* Cleanup HTML using Tidy (or not). * Cleanup HTML using Tidy (or not).
*
* @todo This should be called in init() instead of from __construct
*/ */
private function loadHtml() private function loadHtml()
{ {
@ -266,7 +269,6 @@ class Readability implements LoggerAwareInterface
* Use tidy (if it exists). * Use tidy (if it exists).
* This fixes problems with some sites which would otherwise trouble DOMDocument's HTML parsing. * This fixes problems with some sites which would otherwise trouble DOMDocument's HTML parsing.
* Although sometimes it makes matters worse, which is why there is an option to disable it. * Although sometimes it makes matters worse, which is why there is an option to disable it.
*
*/ */
if ($this->useTidy) { if ($this->useTidy) {
$this->logger->debug('Tidying document'); $this->logger->debug('Tidying document');
@ -314,8 +316,6 @@ class Readability implements LoggerAwareInterface
*/ */
public function init() public function init()
{ {
$this->loadHtml();
if (!isset($this->dom->documentElement)) { if (!isset($this->dom->documentElement)) {
return false; return false;
} }
@ -372,12 +372,31 @@ class Readability implements LoggerAwareInterface
return $this->success; return $this->success;
} }
/**
* Debug.
*
* @deprecated use $this->logger->debug() instead
*/
protected function dbg($msg)
{
$this->logger->debug($msg);
}
/**
* Dump debug info.
*
* @deprecated since Monolog gather log, we don't need it
*/
protected function dump_dbg()
{
}
/** /**
* Run any post-process modifications to article content as necessary. * Run any post-process modifications to article content as necessary.
* *
* @param \DOMElement $articleContent * @param \DOMElement $articleContent
*/ */
public function postProcessContent(\DOMElement $articleContent) public function postProcessContent($articleContent)
{ {
if ($this->convertLinksToFootnotes && !preg_match('/\bwiki/', $this->url)) { if ($this->convertLinksToFootnotes && !preg_match('/\bwiki/', $this->url)) {
$this->addFootnotes($articleContent); $this->addFootnotes($articleContent);
@ -462,7 +481,7 @@ class Readability implements LoggerAwareInterface
* *
* @param \DOMElement $articleContent * @param \DOMElement $articleContent
*/ */
public function addFootnotes(\DOMElement $articleContent) public function addFootnotes($articleContent)
{ {
$footnotesWrapper = $this->dom->createElement('footer'); $footnotesWrapper = $this->dom->createElement('footer');
$footnotesWrapper->setAttribute('class', 'readability-footnotes'); $footnotesWrapper->setAttribute('class', 'readability-footnotes');
@ -526,7 +545,7 @@ class Readability implements LoggerAwareInterface
* *
* @param \DOMElement $articleContent * @param \DOMElement $articleContent
*/ */
public function prepArticle(\DOMElement $articleContent) public function prepArticle($articleContent)
{ {
$this->logger->debug($this->lightClean ? 'Light clean enabled.' : 'Standard clean enabled.'); $this->logger->debug($this->lightClean ? 'Light clean enabled.' : 'Standard clean enabled.');
@ -623,7 +642,7 @@ class Readability implements LoggerAwareInterface
* *
* @param \DOMElement $node * @param \DOMElement $node
*/ */
protected function initializeNode(\DOMElement $node) protected function initializeNode($node)
{ {
if (!isset($node->tagName)) { if (!isset($node->tagName)) {
return; return;
@ -694,7 +713,7 @@ class Readability implements LoggerAwareInterface
* *
* @return \DOMElement|bool * @return \DOMElement|bool
*/ */
protected function grabArticle(\DOMElement $page = null) protected function grabArticle($page = null)
{ {
if (!$page) { if (!$page) {
$page = $this->dom; $page = $this->dom;
@ -743,8 +762,7 @@ class Readability implements LoggerAwareInterface
continue; continue;
} }
// XML_TEXT_NODE if ($childNode->nodeType === XML_TEXT_NODE) {
if ($childNode->nodeType == 3) {
$p = $this->dom->createElement('p'); $p = $this->dom->createElement('p');
$p->innerHTML = $childNode->nodeValue; $p->innerHTML = $childNode->nodeValue;
$p->setAttribute('data-readability-styled', 'true'); $p->setAttribute('data-readability-styled', 'true');
@ -770,7 +788,7 @@ class Readability implements LoggerAwareInterface
continue; continue;
} }
$grandParentNode = ($parentNode->parentNode instanceof \DOMElement) ? $parentNode->parentNode : null; $grandParentNode = $parentNode->parentNode instanceof \DOMElement ? $parentNode->parentNode : null;
$innerText = $this->getInnerText($nodesToScore[$pt]); $innerText = $this->getInnerText($nodesToScore[$pt]);
// If this paragraph is less than MIN_PARAGRAPH_LENGTH (default:20) characters, don't even count it. // If this paragraph is less than MIN_PARAGRAPH_LENGTH (default:20) characters, don't even count it.
@ -1051,7 +1069,7 @@ class Readability implements LoggerAwareInterface
* *
* @return string * @return string
*/ */
public function getInnerText(\DOMElement $e = null, $normalizeSpaces = true, $flattenLines = false) public function getInnerText($e, $normalizeSpaces = true, $flattenLines = false)
{ {
if (null === $e || !isset($e->textContent) || $e->textContent === '') { if (null === $e || !isset($e->textContent) || $e->textContent === '') {
return ''; return '';
@ -1073,7 +1091,7 @@ class Readability implements LoggerAwareInterface
* *
* @param \DOMElement $e * @param \DOMElement $e
*/ */
public function cleanStyles(\DOMElement $e) public function cleanStyles($e)
{ {
if (!is_object($e)) { if (!is_object($e)) {
return; return;
@ -1121,7 +1139,7 @@ class Readability implements LoggerAwareInterface
* *
* @return int * @return int
*/ */
public function getLinkDensity(\DOMElement $e, $excludeExternal = false) public function getLinkDensity($e, $excludeExternal = false)
{ {
$links = $e->getElementsByTagName('a'); $links = $e->getElementsByTagName('a');
$textLength = mb_strlen($this->getInnerText($e, true, true)); $textLength = mb_strlen($this->getInnerText($e, true, true));
@ -1150,7 +1168,7 @@ class Readability implements LoggerAwareInterface
* *
* @return int * @return int
*/ */
protected function weightAttribute(\DOMElement $element, $attribute) protected function weightAttribute($element, $attribute)
{ {
if (!$element->hasAttribute($attribute)) { if (!$element->hasAttribute($attribute)) {
return 0; return 0;
@ -1185,7 +1203,7 @@ class Readability implements LoggerAwareInterface
* *
* @return int * @return int
*/ */
public function getWeight(\DOMElement $e) public function getWeight($e)
{ {
if (!$this->flagIsActive(self::FLAG_WEIGHT_ATTRIBUTES)) { if (!$this->flagIsActive(self::FLAG_WEIGHT_ATTRIBUTES)) {
return 0; return 0;
@ -1205,7 +1223,7 @@ class Readability implements LoggerAwareInterface
* *
* @param \DOMElement $node * @param \DOMElement $node
*/ */
public function killBreaks(\DOMElement $node) public function killBreaks($node)
{ {
$html = $node->innerHTML; $html = $node->innerHTML;
$html = preg_replace($this->regexps['killBreaks'], '<br />', $html); $html = preg_replace($this->regexps['killBreaks'], '<br />', $html);
@ -1221,7 +1239,7 @@ class Readability implements LoggerAwareInterface
* @param \DOMElement $e * @param \DOMElement $e
* @param string $tag * @param string $tag
*/ */
public function clean(\DOMElement $e, $tag) public function clean($e, $tag)
{ {
$currentItem = null; $currentItem = null;
$targetList = $e->getElementsByTagName($tag); $targetList = $e->getElementsByTagName($tag);
@ -1257,7 +1275,7 @@ class Readability implements LoggerAwareInterface
* @param \DOMElement $e * @param \DOMElement $e
* @param string $tag * @param string $tag
*/ */
public function cleanConditionally(\DOMElement $e, $tag) public function cleanConditionally($e, $tag)
{ {
if (!$this->flagIsActive(self::FLAG_CLEAN_CONDITIONALLY)) { if (!$this->flagIsActive(self::FLAG_CLEAN_CONDITIONALLY)) {
return; return;
@ -1370,7 +1388,7 @@ class Readability implements LoggerAwareInterface
* *
* @param \DOMElement $e * @param \DOMElement $e
*/ */
public function cleanHeaders(\DOMElement $e) public function cleanHeaders($e)
{ {
for ($headerIndex = 1; $headerIndex < 3; ++$headerIndex) { for ($headerIndex = 1; $headerIndex < 3; ++$headerIndex) {
$headers = $e->getElementsByTagName('h'.$headerIndex); $headers = $e->getElementsByTagName('h'.$headerIndex);

29
tests/ReadabilityTest.php
Unescape View File

@ -22,40 +22,47 @@ class ReadabilityTest extends \PHPUnit_Framework_TestCase
return $readability; return $readability;
} }
/**
* @requires extension tidy
*/
public function testConstructDefault()
{
$readability = $this->getReadability('');
$this->assertNull($readability->url);
$this->assertInstanceOf('DomDocument', $readability->dom);
}
/**
* @requires extension tidy
*/
public function testConstructSimple() public function testConstructSimple()
{ {
$readability = $this->getReadability('<html/>', 'http://0.0.0.0'); $readability = $this->getReadability('<html/>', 'http://0.0.0.0');
$readability->init();
$this->assertEquals('http://0.0.0.0', $readability->url); $this->assertEquals('http://0.0.0.0', $readability->url);
$this->assertInstanceOf('DomDocument', $readability->dom);
$this->assertEquals('<html/>', $readability->original_html); $this->assertEquals('<html/>', $readability->original_html);
$this->assertTrue($readability->tidied); $this->assertTrue($readability->tidied);
$this->assertTrue($this->logHandler->hasDebugThatContains('Parsing URL: http://0.0.0.0'));
$this->assertTrue($this->logHandler->hasDebugThatContains('Tidying document'));
$this->assertTrue($this->logHandler->hasDebugThatContains('Light clean enabled.'));
} }
public function testConstructDefaultWithoutTidy() public function testConstructDefaultWithoutTidy()
{ {
$readability = $this->getReadability('', null, 'libxml', false); $readability = $this->getReadability('', null, 'libxml', false);
$readability->init();
$this->assertNull($readability->url); $this->assertNull($readability->url);
$this->assertEquals('', $readability->original_html); $this->assertEquals('', $readability->original_html);
$this->assertFalse($readability->tidied); $this->assertFalse($readability->tidied);
$this->assertTrue($this->logHandler->hasDebugThatContains('Parsing URL: ')); $this->assertInstanceOf('DomDocument', $readability->dom);
$this->assertFalse($this->logHandler->hasDebugThatContains('Tidying document'));
$this->assertTrue($this->logHandler->hasDebugThatContains('Light clean enabled.'));
} }
public function testConstructSimpleWithoutTidy() public function testConstructSimpleWithoutTidy()
{ {
$readability = $this->getReadability('<html/>', 'http://0.0.0.0', 'libxml', false); $readability = $this->getReadability('<html/>', 'http://0.0.0.0', 'libxml', false);
$readability->init();
$this->assertEquals('http://0.0.0.0', $readability->url); $this->assertEquals('http://0.0.0.0', $readability->url);
$this->assertInstanceOf('DomDocument', $readability->dom);
$this->assertEquals('<html/>', $readability->original_html); $this->assertEquals('<html/>', $readability->original_html);
$this->assertFalse($readability->tidied); $this->assertFalse($readability->tidied);
} }
@ -447,6 +454,8 @@ class ReadabilityTest extends \PHPUnit_Framework_TestCase
public function testPreFilters() public function testPreFilters()
{ {
$this->markTestSkipped('Won\'t work until loadHtml() is moved in init() instead of __construct()');
$readability = $this->getReadability('<div>'.str_repeat('<p>This <b>is</b> the awesome and WONDERFUL content :)</p>', 7).'</div>', 'http://0.0.0.0'); $readability = $this->getReadability('<div>'.str_repeat('<p>This <b>is</b> the awesome and WONDERFUL content :)</p>', 7).'</div>', 'http://0.0.0.0');
$readability->addPreFilter('!<b[^>]*>(.*?)</b>!is', ''); $readability->addPreFilter('!<b[^>]*>(.*?)</b>!is', '');

Write Preview
Loading…
Cancel
Save